A SYSTEM OF BLOOD ANALYSIS.*

SUPPLEWENT II.

SIMPLIFIED METHOD FOR THE DETERMINATION OF CHLORIDES IN BLOOD OR PLASMA.

BY J. C. WHITEHORN. (8’rom the Biochemical Laboratories of Harvard Medical School, Boston.)

(Received for publication, December 29, 1920.)

The method of preparing protein-free filtrates by the use of tungstic acid, as developed and used by Folin, has proved so highly satisfactory in the determination of a number of blood constituents, that it has seemed desirable to apply it to the deter- mination of chlorides.

The procedure developed for this purpose is based upon the principle of the Volhard method;l namely, the precipitation of silver chloride from a known amount of silver nitrate and titra- tion of the excess silver nitrate by means of sulfocyanate, using ferric ammonium alum as an indicator.

A number of preliminary experiments were conducted to ascer- tain the most convenient conditions which would give a good end-point. Efforts to obtain increased accuracy by greatly increased dilution of reagents were disappointing. In order to insure a sharp end-point, the volume of fluid at the end of titra- tion must be kept small, which cannot be done when very dilute solutions are used. As a result of dilution the end-point tends to be yellowish, rather than red, and cannot be accurately perceived, at least by the writer’s eye, except in strong daylight. This diffi- culty in the use of very dilute solutions is somewhat increased by the presence of oxalate, because of the lemon-yellow color of

* By agreement between Mr. Whitehorn and myself, this paper is pub- lished as Supplement II of the “System of blood analysis” devised by Folin and Wu.-Professor Otto Polin.

1 Volhard, J., Ueber eine neue methode der maassanalytischen Bestim- mung des Silbers, J. prakt. Chem., 1874, ix, 217.

449

by guest on March 4, 2020

http://ww

w.jbc.org/

Dow

nloaded from

450 System of Blood Analysis. II

iron oxalate. It was presumably for this reason that Rappleye’s method for plasma chlorides2 required the use of titrated rather than oxalated plasma. The personal element in the reading of such faint end-points is perhaps rather large, for Myers and Short have very recently published fairly accurate results obtained on picric acid filtrates by a modified Volhard-Arnold method in which they used ammonium thiocyanate of only ~/117 strength.3 The writer, however, has found it more satisfactory to use a thiocyanate solution of ~/35.5 strength, of which 0.03 cc. gives, under the conditions specified, an unmistakable end-point.

For reasons which will appear below, it was necessary to avoid the customary procedure of filtration or centrifugalization of the silver precipitate. This simplification has been accomplished without loss of accuracy by the liberal use of nitric acid and ferric alum. Accurate results were obtained on known chloride solu- tions, with or without the addition of small amounts of tungstic acid.

The following method was thereupon developed. I. Preparation of Protein-Free Filtrates.-The filtrate is pre-

pared by the use of the same reagents as have been described in detail by Folin and Wu4 for the determination of non-protein nitrogen, urea, uric acid, creatinine, creatine, and sugar. Because even slight variations in the chlorides are significant, great accu- racy is necessary. The writer customarily uses volumetric flasks in order to insure an accurate 1: 10 dilution. The method is applicable, without alteration, to either whole blood or plasma. An amount of filtrate equivalent to 1 cc. of blood or plasma is needed. Less may be used but with proportionate loss of accu- racy.

II. Determination of Chloride Content of Filtrate.-(a) Reagents Required.-1. Silver nitrate solution (~/35.46).

2. Potassium (or ammonium) sulfocyanate (~/35.46). 3. Powdered ferric ammonium sulfate (FeNHd(SO&). 4. Concentrated nitric acid (HN03 of specific gravity 1.42).

2 Rappleye, W. C., A simple application of the Volhard principle for blood plasma chlorides, J. Biol. Chem., 1918, xxxv, 509.

3 Myers, V. C., and Short, J. J:, The estimation of chlorides in blood, J. Biol. Chem., 1920, xliv, 47.

4 Folin, O., and Wu, H., A system of blood analysis, J. Biol. Chem., 1919, xxxviii, 81.

by guest on March 4, 2020

http://ww

w.jbc.org/

Dow

nloaded from

J. C. Whitehorn 451

(b) Procedure.-Pipette 10 cc. of the protein-free filtrate into a porcelain dish. Add with a pipette 5 cc. of the standard silver nitrate solution and stir thoroughly. Add about 5 cc. of concen- trated nitric acid, mix, and let stand for 5 minutes, to permit the flocking out of the silver chloride. Then add with a spatula an abundant amount of ferric ammonium sulfate (about 0.3 gm.) and titrate the excess of silver nitrate with the standard sulfo- cyanate solution until the definite salmon-red (not yellow) color of the ferric sulfocyanate persists in spite of stirring for at least 15 seconds.

(c) Cazcuzation.-

5.00 - titer (in cc.) = mg. of Cl per CC. of blood (or plasma)

Since each cc. of thiocyanate solution used is equivalent to 1 cc. of silver nitrate solution, the difference between the volume of silver nitrate solution taken and the excess determined by the titration, that is 5-titer, represents the volume which reacted with chloride at the ratio of 1 cc. to 1 mg. of Cl. And the 10 cc. of filtrate taken represents 1 cc. of blood (or plasma).

To convert Cl figures into NaCl figures divide by 0.606. The same result may be more easily obtained by the following rule: To obtain mg. NaCl per 100 cc., divide mg. Cl per liter by 6, and subtract 0.001 of the result. Conversely, to obtain mg. Cl per liter, add to mg. NaCl per 100 cc. 0.001 of itself and multiply by 6.

The examples in Table I illustrate the principles involved in the calculation.

TABLE I.

I I I NaCl per 100 cc.

Specimen. Cl per cc. By simple rule.

cc. mv. “V. “9.

Filtrate A.. . . . 0.75 (5 - 0.75) = 4.25 2 (4,250) - 7 = 701 701 “ B - = - = . . . . . 2.25 (5 2.25) 2.75 Q (2,750) 5 453 453 “ c..... 1.35 (5 - 1.35) = 3.65 6(3,650) - 6 = 602 602

(d) Preparation of Reagents.-Dissolve 4.791 gm. of C.P. silver nitrate in distilled water. Transfer this solution to a liter volu- metric flask and make up to the mark with distilled water. Mix

by guest on March 4, 2020

http://ww

w.jbc.org/

Dow

nloaded from

452 System of Blood Analysis. II

thoroughly and preserve in a brown bottle. 1 cc. = 1 mg. Cl. (It is to be noted that the silver nitrate and nitric acid are not added to the protein-free filtrate simultaneously. To do so may result in the mechanical enclosure of silver nitrate solution within the curds, and a consequent error in the positive direction.)

Because sulfocyanates are hygroscopic, the standard solution should be prepared volumetrically. As an approximation about 3 gm. of KCNS or 2.5 gm. of NH&NS should be dissolved in a liter of water. By titration under the conditions specified under “Procedure” and by proper dilution prepare a standard such that 5 cc. are equivalent to 5 cc. of the silver nitrate solution.

The solid ferric alum is used rather than a solution, in order to insure a very high concentration in the mixture to be titrated. It is powdered in order to facilitate its solution.

Remarks.

1. Essentially the same procedure and reagents may be used in determining urine chlorides, except for the silver nitrate solu- tion, which should be of M/7.092 strength. When 5 cc. of urine and 5 cc. of this strong silver solution are used in urine chloride

determination the calculation becomes 5 - ‘F = mg. Cl per cc.

of urine. 2. When a determination of both urea and chlorides is desired

on a small sample, as may sometimes occur in cases where nephritis is suspected, one may pipette 2 cc. of the plasma into a 25 cc. flask, dilute to about 20 cc. with water, add 2 cc. each of the tung- state and acid solutions, make up to the mark with water, and shake. This will give sufficient filtrate for both determinations, 5 cc. for urea by Folin’s distillation method and 10 cc. for the chlorides by the method described. In either case the figure obtained must be multiplied by g, since the filtrate has been diluted 2 : 25 instead of 2 : 20.

3. The glassware should be checked to within at least 0.5 per cent.

4. Reagents must be halogen-free. Some samples of nitric acid contain much chloride. None of the samples of tungstate tested has contained chloride. To guard against the possibility

by guest on March 4, 2020

http://ww

w.jbc.org/

Dow

nloaded from

J. C. Whitehorn 453

of contamination, however, all samples should be tested as follows before using: Mix one volume of 10 per cent sodium tungstate solution with two volumes of concentrated, chloride-free nitric acid, and filter into a test-tube containing silver nitrate solution. Turbidity indicates contamination with halogen.

5. At Dr. Folin’s suggestion, purification of tungstates contain- ing added chlorides has been accomplished by recrystallization with alcohol. Sodium tungstate containing 0.3 per cent NaCl was so nearly purified by one recrystallization that the above mentioned silver test gave only a faint opalescence, scarcely perceptible even by transmitted light-not enough to produce an appreciable error. After a second recrystallization no chloride whatever could be detected. The following procedure was used: To a cooled 50 per cent solution of the contaminated tungstate, prepared with the aid of heat, add slowly an equal volume of 95 per cent ethyl alcohol and let stand for 10 minutes. Pour the suspension of crystals on a Buchner funnel, wash twice with alcohol, and dry.

6. The writer’s attention has very recently been called to the method published by Rieger for chloride estimation on tungstic acid filtrates.5 Certain features of this article deserve comment.

(a) The method therein described retains the centrifugalization or filtration procedure, which at times causes erroneously high results on tungstic acid filtrates, whether or not there is sufficient tungstate present to give a precipitate on the addition of an equal volume of concentrated nitric acid. Incidentally, the absence of such a precipitate does not indicate the absence of tungstate, as Rieger has stated, for solutions containing as much as 25 mg. of sodium tungstate per 100 cc. may not give a precipitate with nitric acid except when heated.

(b) The article mentioned contains a method for the “purifica- tion of sodium tungstate;” i.e., for the preparation of chloride- free tungstate. It is very difficult to believe that the method SO designated really accomplishes its purpose, since it calls for the use of 7 cc. of 40 per cent sodium hydroxide solution for each 10 gm. of sodium tungstate taken. Preparations of sodium hydrox- ide always contain large amounts of chloride. On the other

s Rieger, J. B., The estimation of chlorides in whole blood, J. Lab. and Gin. Med., 1920-21, vi, 44.

by guest on March 4, 2020

http://ww

w.jbc.org/

Dow

nloaded from

454 System of Blood Analysis. II

hand, none of the samples of sodium tungstate tested by the writer has contained perceptible amounts of chloride. It has therefore not been necessary to purify. Certainly “purification” should not be attempted by the use of sodium hydroxide.

(c) Instead of precipitating the plasma proteins before dilu- tion it is preferable to dilute the plasma with distilled water before adding the acid, in order to insure an even distribution of chlorides between solution and precipitate.

(cl) It is not necessary to wait an hour for protein precipitation. 5 minutes are sufficient. Time of standing before filtering off the precipitated protein causes no appreciable difference in the chloride determination, as shown by the following results on the same plasma: Stood for 5 minutes before filtering, 3.64 mg. Cl per cc.; 15 minutes, 3.68; 30 minutes, 3.68; 1 hour, 3.64; and 32 hours, 3.70.

Soundness of the Method.

The soundness of the method of chloride determination described obviously depends upon the answers to two questions:

I. Does the titration figure indicate accurately the excess of silver nitrate?

II. Does the amount of silver precipitation so found indicate accurately the chlorides of the plasma?

I. The first of these questions deserves careful attention. Many chemists will quite properly be very skeptical on this point. Sutton state9 “In cases where chlorine is precipitated by excess of silver, and the excess has to be found by thiocyanate, experience has proved that it is absolutely necessary to filter off the chloride and titrate the filtrate and washings.” Rosanoff and Hill’ have shown that the error is due to the reaction of silver chloride with sulfocyanate. Their figures would indicate that silver chloride, when shaken in an equimolar water solution of ammonium sulfocyanate, reacts so rapidly as to precipitate 43 per cent of the sulfocyanate in 2 minutes. This occurs because

6 Sutton, F., Systematic handbook of volumetric analysis, Philadelphia, 10th edition, 1911,145.

7 Rosanoff, M. A., and Hill, A. E., A necessary modification of Volhard’s method for the determination of chlorides, J. Am. 0x-m. Xoc., 1907, xxix, 269.

by guest on March 4, 2020

http://ww

w.jbc.org/

Dow

nloaded from

J. C. Whitehorn 455

silver chloride is more soluble than silver sulfocyanate. Harvey8 has shown, however, that both ferric alum and nitric acid retard this reaction. By using nitric acid in a concentration of 5 per cent, he obtained practically identical results on known chloride solutions and on urines whether or not he filtered off the silver chloride. His results, indeed, indicate a slightly greater accuracy without filtration, but the difference is so small as to be entirely without significance (e.g. 0.7052 without filtration and 0.7069 per cent NaCl with filtration on a known 0.7039 per cent NaCl solution).

The following experiments were designed and carried out to test the accuracy of the titration in the presence of silver chloride. Since the purpose was to discover if the presence of silver chloride influenced the accuracy of the titration, all factors such as tung- state which might modify or conceal such an influence, had to be excluded. Therefore a pure solution of sodium chloride was sub- stituted for the blood filtrate. A 0.1 N solution was prepared by dissolving in distilled water 0.5846 gm. of recrystallized, thor- oughly dried sodium chloride, and making up to 100 cc. in an accurately calibrated flask. Then 1 cc. of this solution and 9 cc. of water’were used instead of 10 cc. of blood filtrate, but all the subsequent details were followed as specified under the heading “Procedure.” Eleven determinations were made in this manner. The chlorine content of 1 cc. of 0.1 M NaCl solution, as deter- mined by these experiments, was: 3.55; 3.52; 3.56; 3.56; 3.54; 3.53; 3.56; 3.55; 3.56; 3.55; and 3.54 mg. The average was 3.547, compared to a theoretical value of 3.546 mg. The highest and lowest deviations were 0.7 and -0.3 per cent. The presence of silver chloride during the titration had therefore produced no error.

Similarly the chlorine content of 4 cc. of 0.1 M NaCl solution, as determined in the same manner (which of course necessitated the use of 3 cc. of M/7.092 AgN03 solution instead of 5 cc. of ~/35.46) in three such experiments, was found to be 14.28, 14.23, and 14.30 mg., compared to the theoretical value of 14.2 mg. Here also the presence of chloride produced no error in titration.

As a final crucial test, another determination was made on 1 cc.

* Harvey, S. C., The quantitative determination of the chlorids in the urine, Arch. Int. Med., 1910, vi, 12.

by guest on March 4, 2020

http://ww

w.jbc.org/

Dow

nloaded from

456 System of Blood Analysis. II

of 0.1 M NaCl solution, but with this difference, that the mixture containing silver chloride was centrifugalized and an aliquot por- tion of the water-clear supernatant Iiquid taken for titration. The whole mixture measured 15 cc. and the titration of 10 cc. of the supernatant liquid required 0.97 cc. of ~/35.46 KCNS solu- tion. Since 5 - i (0.97) = 3.55, it is evident that the method of titration gives the same result whether or not the silver chlo- ride is removed.

The essential part of the procedure is the use of nitric acid, and the principle is a physical one-the flocking out of the silver chloride, with the consequent reduction of the surface exposed for reaction with the sulfocyanate. It is for this purpose that so large an amount of nitric acid is used, approximately 25 per cent of the volume of the mixture in which the titration is carried out.

Much higher concentrations of nitric acid should not be used, as the sulfocyanate is decomposed by them quite rapidly. To determine if such decomposition played an important part in the disappearance of the end-point, a small drop (0.03 cc.) of the M/35.46 KCNS solution was added to each of two dishes, one containing 10 cc. of 30 per cent nitric acid, and the other 10 cc. of 15 per cent nitric acid, and each containing about 0.3 gm. of ferric alum. The red color persisted in the first mixture for 15 minutes, in the second it was still present at the end of 15 hours. It is evident, therefore, that under the conditions which may occur in using this method, the decomposition of sulfo- cyanate by nitric acid is not rapid enough to affect the accuracy of the titration.

In accordance with the geometrical principle that volumes vary as the cubes of a dimension, whereas surfaces vary as the squares, the surface of the silver chloride, when flocked out, is so smaIl that the reaction with ferric sulfocyanate, although prob- ably still going on, is negligible. (The color due to 0.03 cc. of the ~/35.46 KCNS solution persisted for 19 hours in the presence of such curds of silver chloride.) But the small amount of silver chloride which in spite of the nitric acid remains in fine suspen- sion, has a larger aggregate surface and therefore reacts much more rapidly with the sulfocyanate.

Herein lies the explanation of the phenomena observed when KCNS is added to the mixture containing silver nitrate, silver

by guest on March 4, 2020

http://ww

w.jbc.org/

Dow

nloaded from

J. C. Whitehorn 457

chloride, ferric alum, and nitric acid. The red color which appears with each drop disappears in about 3 seconds, because the ferric sulfocyanate reacts with the solution of AgNOB. But when all the silver present as AgN03 has been precipitated, the red color which appears when the next drop is added persists about 30 seconds (from 10 seconds to 1 minute, depending upon how much of the drop was really excess, and upon the thorough- ness of the flocking out of AgCl). Its disappearance is due to the reaction between the sulfocyanate and the very finest of the suspended particles of silver chloride. If another drop is added, the color will persist several minutes, for the sulfocyanate must then react with silver chloride particles of larger size and smaller aggregate surface. If still another drop is added the color will persist from 15 minutes to an hour or more (even for 19 hours),

TABLE II.

Persistence of color (Fe(CNS)s).

C”$kiUg I Excess drops of KCNS solution.

End-point. First. Second. Third.

-- min.

None. 40 sec. 1 90 “ 6 75 “

40 30 min.

4 min. 15 min. 24 “ More than 1 hr. 23 “ “ “ 1 “

More than 1 hr.

More than 1 hr.

for the fine suspension of silver chloride has all been used up, and the aggregate surface left for reaction is very small.

The influence of the silver chloride particles of different sizes can be demonstrated by fractional centrifugalization, as in the experiments given in Table II.

In addition to the action of nitric acid in flocking out the chlo- ride, the abundance of ferric alum used also retards the reaction between silver chloride and sulfocyanate by reducing the ioniza- tion of the latter. This also deepens the end-point color by pre- venting the ionization of the red salt, Fe(CNS)3, into yellow Fe+++ ions and colorless CNS- ions.

II. There remains the second question: “Does the amount of silver precipitation, as found by titration of the excess, indicate accurately t.he chlorides of the plasma?”

by guest on March 4, 2020

http://ww

w.jbc.org/

Dow

nloaded from

458 System of Blood Analysis. II

Mixtures of tungstic acid and chloride bring down more silver than can be accounted for by the chloride alone. This additional precipitation of silver has been found by centrifugalization and titration of the supernatant fluid. Errors as large as 4 per cent were obtained in this way. Similar errors sometimes occurred when centrifugalization was introduced into the chloride deter- mination of tungstic acid filtrates. The amount of error which may so occur is not always proportional to the amount of tungstic acid present, and so appears to, be more probably dependent upon physical than upon chemical reactions.

This possibility of error is, however, entirely avoided in the method described, by carrying on the titration in the presence of the precipitate, when all the silver which has not been truly precipitated by chloride is available for titration. Determina- tions of the same 0.1 M NaCl solution gave practically identical results whether determined directly or after the addition of equal volumes of 10 per cent sodium tungstate solution and 3 N sulfuric acid; for example, with tungstate, 3.53, 3.54, and 3.54 mg. of Cl per cc., and without tungstate, 3.547 as the average of eleven determinations.

The evidence given above has been presented in order to demonstrate that the method is free from error at the two points which seemed, a priori, the most probable sources; namely, the presence of tungstic acid at the time silver chloride is precipitated, and t,he presence of silver chloride at the time of titration.

Checks.

The final test of the accuracy of the method consists of course in its comparison with methods of known accuracy. The check determinations are given in Table III.

These check determinations indicate the essential accuracy of the method described. In order to determine the limit of error, seventeen duplicate determinations, involving nine separate pre- cipitations with tungstic acid, were made on the same plasma, with the following results: 3.56; 3.51; 3.52; 3.54; 3.50; 3.55; 3.57; 3.54; 3.48; 3.49; 3.52; 3.50; 3.52; 3.53; 3.56; 3.54; 3.55 mg. of Cl per cc. The digestion method of von Koranyi showed 3.53 mg. of Cl per cc. The average by the method described is 3.528. The

by guest on March 4, 2020

http://ww

w.jbc.org/

Dow

nloaded from

J. C. Whitehorn

Specimen. N&l checked by method of Cl per cc. NaCl

w. per cent

Whole blood. M

Plasma. M.

2.88 0.473 2.88 0.473 3.69 0.608 3.71 0.612

per cent

Foster.* 0.481 0.479 0.619 0.617

Whole blood. Wh.

Plasma. Wh.

Plasma I. “ II “ 11;. “ IV.

3.06 0.503 3.84 0.633

3.76 0.619 3.78 0.623 3.75 0.618 3.74 0.617

“ v. “ VI “ VI;. “ VIII. “ IX

Whole blood I. Ascitic fluid I.

“ “ II.

3.68 0.606 3.80 0.626 3.52 0.580 3.80 0.626 3.52 0.582 3.24 0.535 4.10 0.676 3.78 0.623

Plasma X. “ XI. ‘I XII. “ XIII. “ XIV

(diluted). Plasma XV.

3.75 0.619 3.38 0.558 3.64 0.600 3.70 0.610

0.504 0.635

McLean-Van S1yke.t 0.6331 0.618$ 0.609$ 0.606$

von Korbnyi.§ 0.603 0.619 0.584 0.631 0.582 0.542 0.677 C.625

Bell and Doisy.11 0.6141 0.567$ 0.5961 0.6051

2.84 0.467 0.465$ 3.86 0.637 0.6411

Average . . . . . . . . . 0.5885 0.5896

* Foster, G. L., A modification of the McLean-Van Slyke method for the determination of chlorides in blood, J. Biol. Chem., 1917, xxxi, 483. (Al- lowance for end-point has been made, as introduced by Van Slyke and Donleavy (Van Slyke, D. D., and Donleavy, J. J., J. Biol. Chem., 1919, xxxvii, 551).

TABLE III.

Method described.

t McLean, F. C., and Van Slyke, D. D., A met,hod for the determination of chlorides in small amounts of body fluids, J. Biol. Chem., 1915, xxi, 361.

$ I am indebted to Mr. L. M. Smith for these determinations. 0 von Koranyi, A., Physiologische und klinische Untersuchungen iiber

den osmotischen Druck thierischer Fliissigkeiten, 2. kli~a. Med., 1897, xxxiii, 1.

/I Bell, R. D., and Doisy, E. A., J. Biol. Chem., 1920-21, xlv, 427.

_- I

by guest on March 4, 2020

http://ww

w.jbc.org/

Dow

nloaded from

460 System of Blood Analysis. II

greatest deviations are - 1.3 and + 1.2 per cent. The limit of error with careful technique must therefore be less than 1.5 per cent.

SUMMARY.

1. A simple and rapid method is described for the determina- tion of bIood and pIasma chlorides.

2. The same reagents as are used for the determination of urine chlorides are employed.

3. The method is especially adapted to the system of blood analysis developed by Folin and Wu.

4. The limit of error is less than 1.5 per cent.

The writer is indebted to Professor Otto Folin for encourage- ment in this work, and to Mr. L. M. Smith for assistance in checking determinations.

by guest on March 4, 2020

http://ww

w.jbc.org/

Dow

nloaded from

J. C. WhitehornCHLORIDES IN BLOOD OR PLASMA

FOR THE DETERMINATION OFSupplement II. SIMPLIFIED METHOD

A SYSTEM OF BLOOD ANALYSIS:

1921, 45:449-460.J. Biol. Chem. 

  http://www.jbc.org/content/45/3/449.citation

Access the most updated version of this article at

 Alerts:

  When a correction for this article is posted• 

When this article is cited• 

alerts to choose from all of JBC's e-mailClick here

  ml#ref-list-1

http://www.jbc.org/content/45/3/449.citation.full.htaccessed free atThis article cites 0 references, 0 of which can be by guest on M

arch 4, 2020http://w

ww

.jbc.org/D

ownloaded from